Back to opsiocloud.com
Skip to registration
Wed 2 Dec 2026, 15:00 IST
Free live webinar for food & beverage manufacturingMake in India

What is your line quietly costing you?

Join Vaishnavi and Praveena from Opsio’s team in Bengaluru for one practical hour on AI-powered visual inspection: where quality losses hide on a food and beverage line, how AI can help, where its limits lie, and how to make the business case on your own line.

Wednesday 2 December

3:00 – 4:00 pm IST, live online

with Vaishnavi & Praveena

Starts in–days–hrs–min

Can’t make it live? Register anyway: the recording and resources follow after the session.

Save your free seat

If a packaging fault started on your line now, when would someone know? (optional, our opening poll)

We use your details to send the joining link and the recording. Privacy policy.

Register and you get

  • Five questions to ask your own plant about false rejects, giveaway, re-inspection and escapes.
  • A plain view of where AI inspection fits, and the places we would not use it.
  • A low-risk way to start: one line, one costly defect, proved before you scale.
  • The recording and resources after the session, even if you can’t make it live.
Vaishnavi ShreeComputer vision expert, presenting
Praveena ShenoyCountry Manager India, moderating
60 minutes, with Q&A

Picture this: the 2:30 check.

Someone finds a smudged date code on a bottle. Most bottles look fine, and perhaps only a few dozen are affected. But which ones, and when did it start? The last good check was at 2:00.

The water is fine and nothing has reached a customer. But people are pulled off other work to sort stock by hand. Nobody would call it a crisis. How many afternoons like this did your plant have last year?

Two water bottles side by side: one with a clear, readable date code, one with the code smudged and unreadableReadableSmudged

Your own line: how much would you have to hold?

Up to

6,000

packs to hold and check, or 500 cases of 12

Your line speed, multiplied by the minutes between checks. The webinar opens here, with a live poll on how long a fault would run on your line.

Register free and bring your numbers
Register free for 2 December

Production is continuous. Inspection often isn’t.

The gap between two checks is where faults run unseen, and where the losses nobody budgets for build up.

Illustrative shift with a check every 30 minutes. A fault that starts just after one check runs until the next.
  • Scrap, rework and false rejects

    Product lost after value has been added, and good product thrown away.

  • Product giveaway

    Small overfills that add up across crores of packs.

  • Re-inspection and stoppages

    Manpower and line time spent re-checking and stopping.

  • Differences between shifts

    Fatigue and personal judgement: the same pack can pass on one shift and fail on another.

  • Complaints, returns and recalls

    Retailer penalties, returned batches and rejected export consignments.

Why it is growing in India: frequent artwork changes, regional and festive packs, SKU-specific MRP and date codes, and export versus domestic packs all mean more ways to get a pack wrong.

Meet your hosts

A computer vision expert who builds inspection AI for factory floors, and the person who runs Opsio in India.

Portrait of Vaishnavi Shree
Presents the session

Vaishnavi Shree

Manufacturing automation & computer vision expert

Director & MLOps Lead, Opsio

  • Computer vision for quality inspection
  • Manufacturing automation
  • MLOps
  • Predictive maintenance
  • Industrial data

Vaishnavi builds and runs the AI that Opsio puts on factory floors: vision models that spot defects on the line, predictive maintenance for industrial plants, and the MLOps that keeps those models accurate after go-live. Her focus is practical AI that holds up in real production, including knowing when not to use it.

Portrait of Praveena Shenoy
Moderates the Q&A

Praveena Shenoy

17+ years bringing technology into Indian operations

Country Manager, India, Opsio

  • AI in manufacturing
  • DevOps
  • Managed services
  • Manufacturing to banking

Praveena leads Opsio in India, with 17+ years across manufacturing, e-commerce, retail, NBFC and banking. Praveena reads your questions out live, makes sure none is too basic to ask, and is the person to call afterwards about starting small on one of your own lines.

The two numbers to ask for

Missed defects and false rejects. A single accuracy figure tells you very little, and both cost you money.

Missed defect

The pack is
Faulty
Inspection says
Good
Your customer finds it, then you pay.

Complaints, returned batches, retailer penalties and rejected export consignments.

False reject

The pack is
Good
Inspection says
Faulty
Your plant pays, every shift.

Good product thrown away or re-checked by hand. Enough false rejects and people stop trusting the inspection.

Where it can earn its money

The visible faults we walk through on the day, each judged by what it costs your plant.

Unreadable date code on a food and beverage packaging lineDate codes & labels
Unreadable date codeMissing or half-printed batch and date codes, caught before dispatch.
Skewed label on a food and beverage packaging lineDate codes & labels
Skewed labelA label wrapped crooked, spotted while the pack is still on the line.
Wrong label on a food and beverage packaging lineDate codes & labels
Wrong labelOne flavour’s label on another’s bottle, after an artwork or SKU change.
Faulty seal on a food and beverage packaging lineSeals & caps
Faulty sealA visibly wrinkled or lifted lid, detected earlier in the line.
Misplaced cap on a food and beverage packaging lineSeals & caps
Misplaced capA cap sitting high or cocked, the tamper ring not engaged.
Fill level on a food and beverage packaging lineFill level
Fill levelUnderfills and giveaway, in containers where the level is visible.

Also in scope: product defects, packaging damage and visible foreign material. Illustrative images.

  1. 1A pack passes

    The camera sees it, the model finds nothing, the line keeps moving.

  2. 2A faulty pack is flagged

    A missing code, a crooked label or a low fill is caught at line speed.

  3. 3The operator sees why

    The defect is marked on screen, so a person can check the decision.

  4. 4The pack is rejected

    The reject mechanism takes it off the line, and the event is logged.

A clear boundary: AI can check visible details such as date, batch and MRP, but compliance with FSSAI and Legal Metrology requirements stays with the manufacturer.

A vision inspection station on a bottling line, with a camera, LED lighting and a reject pusher

How it works

How it works, in plain terms

  1. 1
    Teaching AI what good and bad look like

    The system is trained on images of acceptable and faulty products from your own line.

  2. 2
    Inspecting at production speed

    Cameras, lighting, processing and rejection work as one system.

  3. 3
    Planned for Indian plant conditions

    Heat, humidity, dust, steam and condensation are dealt with at the design stage.

  4. 4
    Not plug-and-play

    Lighting, packaging and product appearance affect performance, so models are validated across SKUs and shifts, monitored and retrained when needed.

Where we would not use AI

An honest part of the hour. Sometimes another tool is better, and sometimes it is simply too early.

  • Hidden contaminants

    Metal detection and X-ray inspection remain the right tools.

  • Seal integrity

    A seal that looks good may still leak, so leak testing stays in place.

  • Opaque containers and net quantity

    Check-weighing is the better fit.

  • Rare defects and poor images

    Without enough usable data, a model will not be reliable.

  • Simple, stable checks

    Conventional machine vision is often enough, and can cost less.

  • When it is too early

    If the cost of the defect is small or unknown, measure it first and invest later.

The hour, in order

An open, experience-led discussion. Questions and your own experiences go in the chat at any time.

  1. The 2:30 check

    A smudged date code, the arithmetic of a fault between two checks, and a live poll: how long before your line would know?

  2. What is your line quietly costing you?

    The inspection gap, false rejects, giveaway, re-inspection, shift differences, complaints, and why it is growing in India.

  3. How it works, in plain terms

    Teaching AI what good and bad look like, inspecting at line speed, Indian plant conditions, and the two numbers to ask for.

  4. Where it can earn its money

    Date codes and labels, visible seal and cap faults, fill level, and what it looks like in use on a line.

  5. Five questions to ask your own plant

    False rejects, giveaway, re-inspection, escapes, and who owns these losses as one number.

  6. Where we would not use AI

    Contaminants, seal integrity, net quantity, rare defects, simple checks, and when it is simply too early.

  7. Starting small, on your own line

    Assess, prove, build the case and scale, and what it means for your inspectors.

  8. Your questions

    Mid-sized plants, existing equipment, defect data, frequent SKU and artwork changes. Praveena reads out questions from the chat.

  9. Closing thoughts

    Back to the opening poll, and three ideas to take back to your team.

Five questions to ask your own plant

Before asking what AI inspection costs, find out what your losses cost.

  1. False rejects: How much good product did you throw away last month, and who in your plant can say for certain?
  2. Giveaway: How much product do you give away free across a year through small overfills?
  3. Re-inspection: How many hours go into re-checking what has already been checked?
  4. Escapes: What did your last complaint or returned batch really cost, once time, freight, credit notes and the relationship are counted?
  5. Ownership: Does anyone own these losses as one number, or do they sit across quality, production and finance?

Starting small

Prove it on your own line

Cameras, lighting, computing, integration, line changes, model upkeep and support all count. Line speed, camera positions, SKUs and defect types drive the cost. Your own line is the proof that matters, from any vendor.

  1. 01Assess

    Choose one line and one high-cost defect, and record what it costs today.

  2. 02Prove

    Run alongside current inspection, against success criteria agreed before the pilot starts.

  3. 03Build the case

    Use your own results, leave out any headcount reduction, and check the decision holds at half the expected savings.

  4. 04Deploy and scale

    Connect to reject systems, define how operators respond, and extend only when the savings hold up.

Your inspectors: their judgement is still needed to review findings, investigate causes and improve the system, and the plan for their roles should be clear from day one.

Colleagues from Opsio’s India team reviewing work together in Bengaluru
Opsio’s India team in Bengaluru
Make in India

Made in India, for Indian lines

The session is run by our team in Bengaluru, the same people who pick up the phone afterwards. It’s timed for Indian plants, and the examples come from the packs on Indian lines: MRP and date codes, regional and festive packs, export and domestic versions of the same product.

The regulatory context: applicable FSSAI and Legal Metrology requirements make accurate labelling, date marking, batch identification and packaged commodity declarations an important part of manufacturing controls. Inspection is where those controls meet the line.

Who it's for

No AI background needed. If you look after what leaves your plant, this hour is for you.

  • Quality heads and QA managers who sign off on what leaves the plant.
  • Production, packaging and process engineers running filling, sealing and labelling lines.
  • Plant heads at beverage, dairy, snacks and packaged food plants, large or mid-sized.
  • Operations and finance leads who want quality losses counted as one number.

Before you register

Does it cost anything?

No. The webinar is free to attend.

I can’t make 2 December. Should I still register?

Yes. Everyone who registers gets the recording and the resources by email after the session.

Where does it happen?

Online, on StreamYard. The joining link is in your confirmation email and calendar invite. No install needed.

Do I need to know anything about AI?

No. It’s built for quality, production and plant people, and it starts from something you already know: what a bad afternoon on the line costs.

Is it a sales pitch?

No. A whole part of the hour is about where we would not use AI, and when it is too early to invest.

Will you cover metal detection or X-ray?

Only to say where they stay the right tool. Hidden contaminants, seal integrity and net quantity need other methods; this hour is about what a camera can see.

Does AI make us compliant with FSSAI or Legal Metrology rules?

No. AI can check visible details such as date, batch and MRP, but compliance with FSSAI and Legal Metrology requirements stays with the manufacturer.

Who do I ask if something isn't covered here?

Praveena Shenoy: praveena.shenoy@opsio.in or +91 63644 60282.

Start with your costliest defect.

Wednesday 2 December 2026, 3:00 pm IST. Free, online, sixty minutes. The recording comes to everyone who registers.

Vaishnavi and Praveena will see you there.

Register free